Analysis of the stride of a walking pedestrian

Abstract

The invention describes a method for analysing the stride of a walking pedestrian, comprising steps of: (a) Acquiring, during a recording period, motion measurements of a lower limb (1) of the pedestrian, (b) Determining each stride made by the pedestrian, (C) Dividing the recording period into recording sub-periods, (d) For several detected strides, estimating a characteristic value of the stride, (e) Selecting at least one characteristic stride value from a predetermined range of percentiles within a set formed by the estimated characteristic stride values, ordered by increasing values and occurring during a recording sub-period, (f) Repeating step (e) for several recording sub-periods. The invention also concerns a piece of equipment (10) for analysing the stride of a walking pedestrian, and a computer program product.

Description

FIELD OF INVENTION

[0001] The present invention relates to the field of gait analysis of a walking pedestrian. TECHNOLOGICAL BACKGROUND

[0002] An individual's gait comprises a sequence of one or more strides. A stride consists of two phases: a first phase during which the foot is in contact with the ground, and a second phase during which the foot is in the air. As illustrated in figure 1 A stride begins when one foot touches the ground, and ends when the same foot touches the ground again.

[0003] Walking requires the use and coordination of several muscles. Therefore, gait analysis can be used to characterize an individual's physical fitness, and in particular the muscular strength they are capable of developing. This muscular strength can vary depending on numerous factors such as the individual's fitness level, age, physical training, or the use of medication that affects muscle efficiency.

[0004] Changes in gait characteristics over time can reflect changes in an individual's physical fitness, particularly in individuals with neuromuscular or neurodegenerative diseases. These diseases, such as Duchenne Muscular Dystrophy (DMD), result in a progressive decline in muscle strength. This leads to altered gait characteristics, with strides typically becoming slower, shorter, and less frequent.

[0005] As is known, an individual's stride can be analyzed during tests, such as the six-minute walk test (6MWT), the North Star Ambulation Assessment (NSAA test), or the timed four-step stair climb test.

[0006] The 6MWT involves asking individuals to walk at their maximum capacity for six minutes, with the distance covered then being analyzed. The 6MWT is used in many fields, such as orthopedics and various neuromuscular pathologies, and aims to measure the maximum effort an individual is capable of producing, in order to characterize their physical fitness level relative to a normative group. Indeed, the maximum power output of an individual is representative of their physical condition and muscle strength.

[0007] The NSAA test includes 17 functional activities, including a 10-minute walk or run, changes between sitting, standing and lying positions, going up and down stairs, jumping, etc.

[0008] Finally, the timed 4-stair climb test measures the minimum time required for an individual to climb four stairs.

[0009] These tests can be repeated at different intervals to monitor changes in an individual's fitness level. However, such tests are demanding because they require being performed in a controlled environment (usually a clinical center), on a specific course (a certain distance, indoors, flat, ideally straight), and in the presence of qualified personnel to ensure the expected conditions are met. The individual must therefore travel regularly to a clinical center to undergo the tests. Furthermore, the individual can only perform these tests during occasional sessions, and the results can be influenced by external factors affecting the individual's performance at the time of testing (temporary fatigue, varying motivation and concentration from one session to another, lane length, surface grip, encouragement given during the test, etc.).During the 6MWT, the individual may, voluntarily or unconsciously, walk faster or slower throughout the tests. The 4-step stair climb test can be very fast, so biases related to the measurement triggering may be observed.

[0010] Therefore, the tests mentioned above may not be representative of an individual's overall fitness level, and the accepted variability for a given individual in these tests, as well as the value corresponding to a clinically substantial change, are very high. For example, for the 6MWT, the accepted variability is approximately 15%, and a clinically substantial change is on the order of 30 minutes.

[0011] As is also well known, some applications, such as pedometers, mobile phones, or smartwatches, can count an individual's steps or the number of activity episodes. The distance covered can then be measured by a GPS receiver built into such devices. However, this measurement lacks precision, and these applications cannot determine the trajectory of a stride. Furthermore, the number of steps is not always representative of an individual's health. Indeed, this parameter is very sensitive to external variables, such as the individual's schedule, any changes in their lifestyle, and so on. Moreover, an individual can take a large number of steps but at a very low speed, or with a very short distance.Finally, a moderate decrease or increase in muscle power may not affect the most common movements, which rarely approach an individual's power or endurance limits. Therefore, simply counting the number of strides, or even providing an average stride length, is insufficient to adequately characterize an individual's fitness level.

[0012] Furthermore, US patent application 2013 / 123665 A1 proposes analyzing an individual's stride trajectory using a device attached to their foot. However, this device does not provide a stride analysis representative of the maximum muscular power an individual is capable of generating, nor does it allow for the analysis of changes in an individual's physical fitness over time.

[0013] Finally, the abstract by SEFERIAN A ET AL: "Longitudinal results of magneto-inertial motion analysis in Duchenne muscular dystrophy ambulant patients", NEUROMUSCULAR DISORDERS, vol. 26, 2016, ISSN: 0960-8966, DOI: 10.1016 / J.NMD.2016.06.357, presents a gait analysis of patients with Duchenne muscular dystrophy. It indicates that, based on ActiMyo® measurements, it is possible to reconstruct the device's trajectory and accurately calculate variables such as speed and step length. It proposes an evaluation based on the density function of step length and speed, calculated over successive intervals of approximately 30 days. SUMMARY OF THE INVENTION

[0014] One aim of the invention is to provide equipment for analyzing the gait of a walking pedestrian that allows tracking the evolution of the pedestrian's physical fitness over time.

[0015] Another aim of the invention is to deduce from a gait analysis information on the maximum muscular power that the individual is capable of developing.

[0016] Another aim of the invention is that this analysis can be carried out continuously, with fewer constraints for the individual.

[0017] The invention is defined by independent claims 1 and 12. Preferred features of the claimed equipment are defined in dependent claims 2-11.

[0018] According to a first unclaimed aspect, a method for analyzing the gait of a walking pedestrian is described, the method comprising the following steps: (a) Acquisition, during a recording period, of lower limb motion measurements of the pedestrian, (b) Determination, by a data processing unit, of each stride taken by the pedestrian during the recording period, based on the motion measurements, (c) Division of the recording period into recording sub-periods, (d) For several detected strides, estimation of a characteristic stride quantity, based on the motion measurements, in which the characteristic quantity is or depends on a stride length, (e) Selection of at least one characteristic stride quantity within a predetermined percentile range from a set of estimated characteristic stride quantities, ordered by increasing values ​​and occurring during a recording sub-period, (f) Repetition of step (e) for several recording sub-periods.

[0019] Selecting, during step (e), a characteristic stride quantity within a predetermined percentile range, between the 70th and 100th percentile, for a given recording sub-period, provides a simple and relevant indicator (in this case, a percentile of this characteristic quantity) of an individual's physical fitness during a given sub-period.

[0020] Repeating step (e) during step (f) for several sub-periods of recording allows us to track the evolution of an individual's physical fitness over time.

[0021] A high percentile of characteristic magnitude reflects the maximum muscular power that the individual is capable of developing.

[0022] Some preferred characteristics of the process are as follows, taken individually or in combination: The characteristic quantity of a stride is its average stride speed. The predetermined percentile range consists of the 95th percentile. Motion measurements are acquired using at least one inertial sensor, such as an accelerometer or gyroscope. The inertial sensor is attached to the pedestrian's lower limb, for example, at the ankle. This arrangement offers advantages in terms of ergonomics, comfort, aesthetics (sensor concealed under trousers), and safety (risk of falling if a foot-mounted sensor catches on an external object), particularly compared to a sensor attached to the foot. Stride determination involves detecting, at a given instant, an acceleration with an absolute value exceeding a predetermined threshold; this instant defines the beginning of the current stride and the end of a previous stride.Stride length is the length of the projection of a stride path onto a horizontal plane, or the curvilinear length of a stride path, said path being estimated based on movement measurements. Average stride speed is the ratio of stride length to the time elapsed between its beginning and end. The stride analysis procedure further includes a step to verify a criterion for a stride belonging to a continuous sequence of several consecutive strides, steps (d) to (f) of the procedure being performed only on strides belonging to a continuous sequence of several consecutive strides. Such a step makes it possible to analyze only those strides representative of a natural and continuous gait of the individual, without the analysis being biased by shuffling, isolated strides, or small movements.A stride belongs to a continuous sequence of several consecutive strides when the time between the start of said stride and the start of a preceding or following stride is between 0 and 10 seconds, preferably less than 3 seconds, and said stride is part of a sequence of at least two consecutive strides, preferably at least six consecutive strides. A recording sub-period has a duration greater than 2 days, preferably greater than 15 days, and / or a cumulative recording duration greater than 5 hours, 10 hours, or 50 hours, preferably greater than 180 hours. These criteria make it possible to define a sub-period containing a sufficient amount of data to be representative of the individual's daily activity, without the individual's fitness level being likely to change significantly during this period.

[0023] According to a second aspect, claimed by independent claim 1, the invention relates to equipment for analyzing the gait of a walking pedestrian, the equipment comprising: At least one inertial sensor to acquire, during a recording period, motion measurements of a lower limb of the pedestrian, A data processing unit configured to: o Determine each stride taken by the pedestrian during the recording period, based on the motion measurements, o Divide the recording period into sub-periods, o For several detected strides, estimate a characteristic stride quantity, based on the motion measurements, the characteristic quantity being or dependent on a stride length, o Select at least one characteristic stride quantity in a predetermined percentile range between the 70th and 100th percentiles from a set formed by the estimated characteristic stride quantities, ordered by increasing values ​​and occurring during one of the sub-periods, o Repeat the previous step for several sub-periods.

[0024] According to a third aspect, claimed by independent claim 12, the invention relates to a computer program product comprising code instructions for the execution of a method for analyzing the stride of a walking pedestrian according to the first aspect when this program is executed by a processor. PRESENTATION OF THE FIGURES

[0025] Other aspects, purposes and advantages of the present invention will become apparent from the detailed description that follows, given by way of non-limiting example, which will be illustrated by the following figures: There figure 1 The diagram, already discussed, represents the strides of an individual. figure 2 is a diagram representing a pedestrian gait analysis device conforming to the invention. figure 3 is a diagram representing a pedestrian gait analysis device conforming to the invention, attached to a pedestrian's lower limb. figure 4is a graph representing an example of the trajectory of an ankle over several strides. DETAILED DESCRIPTION OF A METHOD OF IMPLEMENTATION Gait analysis equipment

[0026] With reference to the figure 2 A gait analysis device 10 includes inertial sensors 11. The inertial sensors 11 may include an accelerometer and a gyroscope. Preferably, the inertial sensors 11 include an inertial measurement unit comprising at least three accelerometers and three gyroscopes, for example, of the MEMS type. The gait analysis device 10 may also include a battery, a magnetometer 14, an altimeter, and / or a GPS.

[0027] The gait analysis equipment 10 may include a data processing unit 12 (typically a processor) for real-time data processing. The gait analysis equipment 10 may also include storage means 13 (e.g., flash memory) capable of storing measurements taken and / or data to be processed or processed by the data processing unit 12. The gait analysis equipment 10 may also include communication means 15 for transmitting an estimated position. For example, the wearer's position may be sent to a mobile terminal for display in a navigation software interface.

[0028] Alternatively, the gait analysis equipment 10 may include communication means 15 for transmitting measurements to an external device such as a mobile terminal or a remote server. These communication means 15 may implement short-range wireless communication, for example Bluetooth or Wi-Fi (particularly in an embodiment with a mobile terminal), or even be means of connection to a mobile network (typically UMTS / LTE) for long-distance communication. It should be noted that the communication means 15 may, for example, be a wired connection (typically USB) for transferring data from local data storage means 13 to those of a mobile terminal or server. The mobile terminal or remote server then includes a data processing unit 12 (typically a processor) for implementing the data processing.

[0029] In the remainder of this description, we will see that the data processing unit 12, respectively of the stride analysis equipment 10, of a mobile terminal and of a remote server, can indifferently and according to the applications carry out all or part of the steps of the process.

[0030] The stride analysis equipment 10 may further include a housing 17 comprising the inertial sensors 11. The housing 17 may further include the optional data processing unit 12, storage means 13 and communication means 15.

[0031] With reference to the figure 3 The pedestrian has at least one lower limb 1, such as a leg. It will be understood that one or both of the pedestrian's lower limbs 1 may be equipped with gait analysis equipment 10.

[0032] The gait analysis device 10 may further include attachment means 16 to the lower limb 1, for example, to the foot, ankle, shin, or thigh of the pedestrian. For example, the gait analysis device 10 includes attachment means 16 at the ankle 3 of the pedestrian. The attachment means 16 of the unit 17 may include a strap passed around the lower limb 1, or a hook-and-loop fastener that secures the lower limb 1 and allows it to be firmly attached to the unit 17 (and thus to the inertial sensors 11 it contains). In this way, the inertial sensors 11 exhibit a movement substantially identical in the Earth's frame of reference to the movement of the pedestrian's ankle 3. Furthermore, the unit 17 and its attachment means 16 can remain concealed and protected under the individual's trousers. The individual can wear the 17 case daily and continuously, which increases the relevance of the gait analysis.However, these inertial sensors 11 can also be placed, for example, on the pedestrian's shin or thigh. Gait analysis method

[0033] A method for analyzing the gait of a walking pedestrian includes the following steps.

[0034] The inertial sensor 11 is preferably attached to the pedestrian's lower limb 1, for example to their foot, ankle, shin, or thigh. Preferably, the inertial sensor 11 is attached to the pedestrian's ankle 3.

[0035] The gait analysis method includes a first step (a) of acquiring, during a recording period, motion measurements of a lower limb 1 of the pedestrian. The acquisition is preferably carried out by at least one inertial sensor 11, such as an accelerometer or a gyroscope. The motion measurements are, for example, measurements of the acceleration and angular velocity of the pedestrian's ankle 3.

[0036] The recording period corresponds to the duration for which the individual is likely to wear the gait analysis device 10, during which time the inertial sensors 11 can perform inertial measurements. For example, an individual might wear the device 17 during the day, approximately twelve hours per day. In this case, for a 15-day recording period, the gait analysis device 10 will have recorded 180 hours of data. It is also possible that the individual might wear the gait analysis device 10 for twelve hours per day, but only on certain days, for example, every other day. In this case, for the same 15-day recording period, the gait analysis device 10 will have recorded only 90 hours of data.

[0037] The gait analysis process includes a second step (b) in which the data processing unit 12 determines each stride taken by the pedestrian during the recording period, based on the motion measurements. This involves determining that a stride has been taken, as well as determining the start time and the end time of the stride.

[0038] In the case of a gait analysis device attached to a pedestrian's shoe, gait determination can be performed by searching for phases of zero angular velocity and acceleration magnitude equal to gravity. This method is known in particular within the framework of ZUPT (Zero velocity update) methods for example [Foxlin, Eric, "Pedestrian Tracking with Shoe-Mounted Inertial Sensors", IEEE Computer Society, Nov. / Dec. 2005, pp. 38-46].

[0039] In the case of a gait analysis device 10 located on the ankle 3, the measured acceleration and angular velocity are never zero, even when foot 2 is on the ground, because there is rotation around the heel during the phase when foot 2 is in contact with the ground. Therefore, methods such as the ZUPT method cannot be directly applied, and it is preferable to rely on criteria other than zero or near-zero measurements to determine a gait pattern.

[0040] In a preferred embodiment, stride determination involves detecting, at a given instant, an acceleration with an absolute value exceeding a predetermined threshold. This instant defines the beginning of the current stride and the end of the previous stride. Indeed, at the moment of impact of foot 2 on the ground, a shock occurs throughout the lower limb 1. The absolute value of the acceleration measured by the accelerometer is then likely to exceed the predetermined threshold. This threshold could, for example, be between 5 m / s² and 50 m / s², or be equal to 15 m / s².

[0041] To improve the accuracy of stride detection, it is possible to combine the detection of an absolute acceleration exceeding a predetermined threshold with a second stride determination criterion. This second criterion can depend on the time elapsed since impact, the specific acceleration measured over a certain period, the acceleration peak corresponding to the impact, the angular velocity measured over a period elapsed since impact, and / or the change in angular velocity over a period elapsed since impact.

[0042] More precisely, this second criterion can take the form of minimizing an expression that depends on one or more of the parameters mentioned above. The minimum of the expression can be sought within a certain time after the impact, for example, a time between 0 seconds and 1 second after the impact, preferably 0.5 seconds after the impact. The expression to be minimized can be written in the form: Δt choc + accel X ms − Choc + α ∗ Vang X ms + β ∗ ΔVang X ms , Or Shock represents the value of the last detected specific acceleration peak, i.e., having a norm greater than a predetermined threshold of 1g, preferably greater than 1.05g, Δt shock represents the time elapsed between the detected shock and the current moment by the acceleration with an absolute value greater than the predetermined threshold, | accel| represents the average absolute value of the specific acceleration measured in m / s² during X ms, X preferably being between 10 and 500, preferably equal to 40, α is a number, for example between 1 and 100, preferably equal to 12, | Vang | X ms represents the average of the absolute value of the angular velocity measured in ° / s over X ms, X preferably being between 10 and 500, preferably equal to 40, β is a number, for example between 1 and 100, preferably equal to 10, ΔVang X ms represents the variation of the angular velocity in the axis perpendicular to the leg and foot and measured in ° / s during X ms, X being preferably between 10 and 500, preferably equal to 40.

[0043] The start of a stride can be defined as the instant the expression below is minimized, this instant also corresponding to the end of the previous stride. The combination of detecting an acceleration with an absolute value greater than a predetermined threshold and minimizing the above expression allows for the precise and consistent determination of a stride.

[0044] Alternatively, machine learning methods such as a neural network, a random forest, regression or any other known statistical learning algorithm, can be used to identify a typical pattern of measurement variations, this pattern making it possible to determine a stride.

[0045] The gait analysis process includes a third step (c) of dividing the recording period into sub-periods. This step allows for tracking the evolution of the measured variables, and therefore an individual's fitness level, sub-period by sub-period. For example, if the recording period is six months, it may be useful to have the evolution of the individual's fitness level month by month, with the recording period divided into six one-month sub-periods, rather than having a single result for the entire recording period.

[0046] A recording sub-period must have a duration and cumulative recording duration sufficiently long so that the gait analysis it contains is representative of an individual's fitness level. In a preferred embodiment, a recording sub-period has a duration greater than 2 days, preferably greater than 15 days, and / or a cumulative recording duration greater than 5 hours, 10 hours, or 50 hours, preferably greater than 180 hours.

[0047] Indeed, if an individual did not wear the gait analysis device 10 for 15 days, the corresponding recording sub-period would last 15 days, but with a cumulative recording time of 0 hours. Such a sub-period would contain no strides and would therefore not be representative of an individual's fitness level.

[0048] A recording period of 15 days and 180 hours is, a priori, long enough to guarantee a sufficient amount of data to be representative of the individual's daily activity, while being short enough that the individual's fitness level is not significantly altered during this period. Such a duration also reduces the variability of the measurements, without placing too much strain on the individual wearing the gait analysis equipment.10

[0049] Alternatively, only one or the other of these criteria of duration and cumulative duration of recording of the sub-period must be met.

[0050] The gait analysis process includes a fourth step (d) which consists of estimating a characteristic stride quantity for several detected strides, based on the movement measurements. The estimated characteristic quantity could, for example, be an average stride speed or a stride length.

[0051] Preferably, the data processing unit 12 estimates the trajectory of ankle 3 during the recording period. This estimation can be performed using a double integration of the motion measurements, preferably recalibrating the double integration each time foot 2 touches the ground. A device for estimating an ankle trajectory is described in document FR 3 042 266. The figure 4 represents an example of trajectory representation estimated over three consecutive strides.

[0052] According to a preferred embodiment, stride length is the length of the projection of the stride trajectory onto a horizontal plane (i.e., a plane parallel to the ground on which the individual wearing the gait analysis equipment is walking), the trajectory being estimated based on movement measurements, as described above. Alternatively, stride length can be, for example, the curvilinear length of the estimated trajectory, without projecting it onto the horizontal plane.

[0053] The average speed of a stride can be defined by the ratio between the length of the stride and its duration, the duration of a stride corresponding to the time elapsed between its beginning and its end.

[0054] Determining an average stride speed offers several advantages over continuously measuring instantaneous stride speed during walking or sampling at high frequencies. Very slow strides, which are longer than fast strides, would then be associated with a greater number of speeds compared to faster strides, thus skewing the results. Furthermore, deriving the trajectory used to estimate instantaneous speed introduces noise that could affect the estimated instantaneous speed values. Finally, assigning a single average speed to a given stride simplifies data processing by reducing the amount of data required for subsequent statistical calculations.

[0055] Another embodiment consists of considering the velocity along each trajectory at the sampling frequency allowed by the calculation, in a preferred embodiment between 50Hz and 1000hz, preferably in the vicinity of 100Hz.

[0056] The stride analysis process includes a fifth step (e) of selecting at least one characteristic stride quantity (e.g., average speed and / or length) within a predetermined percentile range from a set of estimated characteristic stride quantities, ordered by increasing values ​​and occurring during a recording sub-period.

[0057] A stride can be associated with its date, time, and / or start and / or end times, allowing it to be linked to a specific recording sub-period. A stride can also be associated with its characteristic value (e.g., average speed and / or length). The strides within a given sub-period can be ordered by increasing values ​​of these characteristic values.

[0058] A statistical calculation can then be performed on a set of estimated stride characteristics (e.g., average speeds and / or lengths), ordered in ascending order, that occurred during a recording sub-period. Specifically, it is possible to calculate a percentile of this characteristic relative to all stride characteristics within the recording sub-period. For example, it is possible to calculate the 50th percentile (the median), the 80th percentile, the 95th percentile, or any other percentile value of the stride characteristic for a given sub-period.

[0059] A predetermined range of percentiles can be defined, this range being of interest for gait analysis. In particular, the higher percentiles are representative of the maximum effort and maximum muscular power an individual is capable of developing, as they reflect an individual's fastest and / or longest strides, walking speed being constrained by muscular strength. These higher percentiles are therefore particularly sensitive to an individual's fitness level. The term "high percentile" is used in this application to refer to a percentile above the 70th percentile.

[0060] This type of gait analysis, based on a range of high percentiles, is particularly well-suited to walking, as individuals statistically take a number of longer and / or faster strides in their daily lives. Specifically, the few percent of longer and / or faster strides are representative of the maximum power an individual is capable of generating. For example, studies have found a correlation between measurements from exercise tests such as the 6MWT and the 95th percentile of stride length and / or speed in the same individual.

[0061] The predetermined percentile range is between the 70th and 100th percentiles. The predetermined percentile range may include the 95th percentile.

[0062] The gait analysis process includes a fifth step (f) of repeating step (e) for several sub-periods of recording.

[0063] Such a step (f) allows access to characteristic stride lengths within a predetermined percentile range for several recording sub-periods. The predetermined percentile range is the same for each sub-period for which step (e) is repeated. Thus, it is possible to track the evolution of the characteristic stride length, sub-period after sub-period. The evolution of the fitness level over time of the individual wearing the gait analysis equipment 10 can then be monitored.

[0064] In a preferred embodiment, the stride analysis method further includes a step of verifying a criterion of belonging of a stride to a continuous sequence of several consecutive strides, steps (d) to (f) of the method being carried out only on strides belonging to a continuous sequence of several consecutive strides.

[0065] This step allows for the analysis of data derived solely from continuous, natural walking, when the individual is in motion. Therefore, the data is not at risk of being biased by a significant number of shuffles, isolated strides, or micro-movements, as these are not representative of the individual's fitness level. Consequently, the strides considered for calculating the predetermined percentile range, particularly the 95th percentile, do not include isolated strides or shuffles, thus eliminating the corresponding measurement bias.

[0066] Preferably, a stride belongs to a continuous sequence of several consecutive strides when the time between the start of said stride and the start of a previous or following stride is between 0 and 10 seconds, preferably less than 3 seconds, and said stride is part of a sequence of at least two consecutive strides, preferably of at least six consecutive strides.

[0067] Indeed, a duration between two strides exceeding, for example, 0.5 seconds, 3 seconds, or 10 seconds, indicates that the individual likely stopped between strides. The strides then correspond to shuffling, do not belong to a single walking sequence, and do not reflect the individual's fitness level.

[0068] Furthermore, during a sequence of less than, for example, ten, six, or two consecutive strides, the individual is not a priori in a normal walking rhythm, but in a slow rhythm, corresponding to shuffling or a small movement.

[0069] The combination of these two criteria makes it possible to exclude strides that are not characteristic of continuous walking by the individual and the muscular power they are capable of developing. Alternatively, only one of the two criteria above needs to be met for steps (d) to (f) of the process to be carried out.

[0070] With the same objective of processing only strides representative of an individual's fitness level, it is possible to identify strides containing errors, for example, in trajectory calculation or in determining their start and end points. To this end, a stride trajectory can be compared to one or more reference trajectories. The shape of the trajectory can be studied, the existence of a maximum vertical position around the midpoint of the stride can be verified, and ratios between trajectory height and length can be defined. Strides that deviate excessively from the expected characteristics based on the reference trajectories are then identified as erroneous, and steps (d) to (f) of the process are not performed on these strides. Thus, the data analysis is not biased by potential errors in trajectory calculation, stride determination, or other factors.

[0071] Alternatively, a machine learning algorithm such as a neural network, a random forest, an SVM or any other known method, can be used to compare a stride to a database of reference strides, and thus identify an erroneous stride.

[0072] Furthermore, it is possible to identify the first and last strides of each step, as steps (d) to (f) of the process are not performed on these strides. Indeed, these strides are not representative of a continuous walk by an individual.

[0073] It is also possible to identify strides involving a change of direction, as steps (d) to (f) of the process are not performed on these strides. Thus, only strides belonging to a straight-line walk are retained. Indeed, a change of direction causes deceleration, and a stride involving a change of direction is therefore not representative of a continuous, running walk by an individual. Moreover, such a stride is likely to be less well-defined, as the rotation and acceleration pattern of the stride is altered by the change of direction.

[0074] Step (a) of the stride analysis process can be implemented by at least one inertial sensor 11 of the stride analysis equipment 10.

[0075] Steps (b) to (f) of the stride analysis process, as well as the potential additional steps described above, can be implemented by the data processing unit 12 of the stride analysis equipment 10.

Claims

1. Equipment (10) for analyzing the stride of a walking pedestrian, the equipment (10) comprising: - At least one inertial sensor (11) to acquire, during a recording period, measurements of a motion of a lower limb (1) of the pedestrian, - A data processing unit (12) configured to: o Determine each stride made by the pedestrian during the recording period, based on the motion measurements, o Divide the recording period into sub-periods, o For several detected strides, estimate a characteristic quantity of the stride, based on the motion measurements, the characteristic quantity being or depending on a stride length, o Select at least one characteristic stride quantity in a predetermined range of percentiles comprised between the 70th and 100th percentile within a set formed by the estimated characteristic stride quantities, ordered by increasing values and occurring during one of the sub-periods, o Repeat the previous step for several sub-periods.

2. The equipment (10) according to the preceding claim, wherein the characteristic quantity of a stride is an average velocity of the stride.

3. The equipment (10) according to any of the preceding claims, wherein the predetermined range of percentiles consists of the 95th percentile.

4. The equipment (10) according to any of the preceding claims, wherein the inertial sensor (11) comprises an accelerometer and / or a gyrometer.

5. The equipment (10) according to any of the preceding claims, wherein the inertial sensor (11) is configured to be attached to the lower limb (1) of the pedestrian, for example at an ankle (3) of the pedestrian.

6. The equipment (10) according to any of the preceding claims, wherein the determination of a stride comprises the detection for a given moment of an acceleration of absolute value greater than a predetermined threshold, this moment defining the start of the stride and the end of a previous stride.

7. The equipment (10) according to any of the preceding claims, wherein the length of a stride is the length of the projection of a path of the stride on a horizontal plane, or is the curvilinear length of a path of the stride, said path being estimated based on the motion measurements.

8. The equipment (10) according to claim 6 as dependent on claim 2, wherein the average velocity of a stride is a ratio between the length of the stride and the duration elapsed between its start and its end.

9. The equipment (10) according to any of the preceding claims, wherein the data processing unit (12) is further configured to verify a criterion of belonging of a stride to a continuous sequence of several consecutive strides, and to implement the steps of estimation, selection and repetition only on the strides belonging to a continuous sequence of several consecutive strides.

10. The equipment (10) according to claim 9, wherein a stride belongs to a continuous sequence of several consecutive strides when the duration between the start of said stride and the start of a previous or next stride is comprised between 0 and 10 seconds, preferably less than 3 seconds, and when said stride is part of a sequence of at least two consecutive strides, preferably at least six consecutive strides.

11. The equipment (10) according to any of the preceding claims, wherein a recording sub-period has a duration greater than 2 days, preferably greater than 15 days and / or a cumulative recording duration greater than 5 hours, 10 hours or 50 hours, preferably greater than 180 hours.

12. A computer program product comprising code instructions for the execution, when said program is executed on a computer, of a method for analyzing the stride of a walking pedestrian comprising steps of: (a)Receiving measurements of a motion of a lower limb (1) of the pedestrian acquired during a recording period, (b) Determining, by a data processing unit (12), each stride made by the pedestrian during the recording period, based on the motion measurements, (c) Dividing the recording period into recording sub-periods, (d)For several detected strides, estimating a characteristic quantity of the stride, based on the motion measurements, in which the characteristic quantity is or depends on a stride length, (e) Selecting at least one characteristic stride quantity in a predetermined range of percentiles within a set formed by the estimated characteristic quantities, ordered by increasing values and occurring during a recording sub-period, the predetermined range of percentiles being comprised between the 70th and 100th percentile, (f) Repeating step (e) for several recording sub-periods.

Images